New
WEBINAR | AI for Marketers: Content Generation and Personalization with Visual Search
March 8, 2019

How Computer Vision Can Improve Buyer Conversion and Retention for e-commerce sites

Table of Contents:

Converting visitors to buyers then buyers to customers are two of the most critical challenges e-commerce businesses must overcome. While offering quality products and exceptional customer service will remain at the core of achieving these goals, companies can now optimize both by adding one particular artificial intelligence technology to their platform: computer vision.

Here are 3 ways computer vision can help your business to earn and retain loyal customers:

 

Use computer vision to learn what your target customers really want to buy

 

While social media can be a goldmine of information for retailers, you’ll need the right tool to take advantage of it. Through the millions of posts from members of your target audiences, you can often find out what these buyers like or want from the products they buy and use. With computer vision’s ability to quickly process large amounts of visual data, it can be the perfect data analysis tool for mining this repository.

henri-meilhac-134082-unsplash (1)

When Pulsar, a social-listening platform, wanted to see how two different key audiences were wearing Vogue magazine’s predicted Spring/Summer 2019 trends, they turned to Clarifai’s visual recognition capabilities. With computer vision, Pulsar was able to process millions of social media pictures from fashion week posted by professional fashionistas and the street style crowd. From there, they were able to see that some of Vogue’s predicted trends–bows, beige, craft, cycle shorts, actually had little to no adoption among either audience. Further, even where both audiences adopted the same trends like lace and hats, each group took these trends in different directions. Imagine being able to quickly discern not only what trends are really making an impact among potential buyers but the kind of impact they’re having.

Having this data at their fingertips would be invaluable to online apparel retailers, as these companies could curate their inventory even more accurately, ensuring they meet the current tastes of their target market. This, in turn, could help them to attract new buyers to their brand who will then stick around because they see their tastes are being catered to.

 

Use computer vision to let customers call it whatever they like and still find your product

 

While ensuring your inventory meets the target market’s preferences is very important, e-commerce businesses won’t see returns there if these buyers can’t find their products. To make sure their products are discovered, e-commerce companies know they must use the right keywords in their product tags and metadata. Still, this is no simple feat. While buyers may follow the same trends, they don’t always use the same vernacular to describe them.

This is another reason why visual recognition is a worthwhile investment. You can quickly train computer vision models to see and apply many different concepts or tags to the same object, capturing all the search terms buyers may use when looking for products to buy (even in other languages!) By using computer vision to tag these images, you can quickly make your product tags and metadata all-encompassing, so customers can easily and organically find you. For instance, check out the below image.

null

 

Taken from a search demo we built for Clarifai customer, West Elm, you can see we labeled this particular lighting fixture as “modern,” as that is the style category it fits into for West Elm. If a buyer uses that same jargon, searching for “modern light fixture” in a third-party search engine, this tag makes sure West Elm’s product surfaces in the search result, as shown in the below.

Screen Shot 2019-03-06 at 3.37.27 PM

 

Still, not every buyer will use such specialized language. This is why this custom demo was built on top of Clarifai’s General model. The general model applies more generic tags to this product image’s tags and metadata. You can see them on the right in the next image.

WestElmGeneralSearch

 

As a result, even where a customer doesn’t see style the way West Elm does or uses broader search terms, like “gold,” “hanging,” or “lamp,” in that third-party search engine, that buyer will still find West Elm’s product offering.

WestElmGoogleSearch

By tagging their product comprehensively, West Elm is ensuring the right product is found by the right customer, no matter what jargon they use.

 

Want your own custom demo?

Talk to us!

 

 

Use computer vision to add visual search to your site

 

Since humans are visual creatures, we process images far better than words. It’s, therefore, often much easier for a customer to use visual inputs as a search query than to try to describe that item in words. With this in mind, the use of visual search by retail customers has grown immensely over the last year. Adding comprehensive metadata will ensure potential buyers find your products using a third-party search engine that allows for visual search. Still, what happens after they go to your store? Visual recognition companies like Clarifai can help businesses integrate this capability into your internal search engine, giving the buyers who are already on your site a highly efficient way to find the exact product what they want within your catalog, without even having to leave.

Retailers who invest in visual search can expect to see significant returns on that investment. While Pinterest reported that home decor and retail were the two largest retail categories for which their customers used visual search, another study predicted that any retailer who redesigned their websites to allow for any visual search could see their eCommerce revenue increase by 30% by 2020.

architizer_3_screenshot

Architizer, for instance, is an online platform that connects architects with the building products they need. As their customers heavily rely on visual content to make buying decisions, the company needed to add visual search to its platform. Using Clarifai’s visual recognition capabilities, Architizer trained a custom model to detect patterns and visual similarities in the broad array of architecture-themed photos on their site. Buyers could then take an image that inspired them and use it to find visually similar content and connect with the architects and manufacturers behind the designs they wanted to emulate.

Visual search also allows you to show off your inventory. West Elm’s Pinterest Stylefinder tool, also built with Clarifai, uses buyer’s Pinterest boards to find items that are visually similar across their many product categories. This means not only will the tool find the throw blanket a customer is looking for it can also show them a chair cushion they hadn’t even thought of that matches with their desired style, potentially increasing their basket size and showing customers they can get everything they need to decorate from West Elm.

  null

 

Check out Clarifai's interactive Search for Retail demo "search by image" and "search by multiple concepts or images" features in action. 

 
With computer vision, online retailers can optimize the investments they’ve already made to give their customers the best products, by helping them to attract the right customers for those products and encouraging these buyers to stick around.

 

 The Retailer's Visual Recognition Cheat Sheet